Apg7p/Cvt2p: A Novel Protein-activating Enzyme Essential for Autophagy

INTRODUCTION

Autophagy is the process of bulk degradation of cytoplasmic components by the lysosomal/vacuolar system (Seglen and Bohley, 1992;Dunn, 1994). The phenomenon is dramatically enhanced under nutrient starvation conditions. In the initial step of the macroautophagy, a cup-shaped membrane sac surrounds cytosolic components to form an autophagosome (Baba et al., 1994). The outer membrane of the autophagosome fuses with a lysosome/vacuole (Baba et al., 1995). A transient single-membrane structure, the autophagic body, is released into the lumen and subsequently degraded in the lysosome/vacuole. Although biochemical and cell–biological approaches have revealed several aspects of autophagy in mammalian cells, the molecular mechanism of autophagy remains unclear.

In the yeast Saccharomyces cerevisiae, macroautophagy was first described by Takeshige et al. (1992), and the process is similar to that in higher eukaryotes (Baba et al., 1994,1995). Taking advantage of yeast genetics, autophagy-defective mutants (14 apg mutants and 9 aut mutants) have been isolated in two different laboratories (Tsukada and Ohsumi, 1993; Thummet al., 1994; Harding et al., 1996). Surprisingly, most of the apg and aut mutants overlap genetically and phenotypically with some cvtmutants, which have defects in the cytoplasm-to-vacuole targeting of aminopeptidase I (API)¹. These results suggest that the APG and AUT gene products function even in a vegetative growth (APG1/CVT10/AUT3, APG7/CVT2, APG8/CVT5/AUT7, APG9/CVT7/AUT9, APG14/CVT12, APG15/CVT11, andCVT17/AUT5) (Harding et al., 1996; Scott et al., 1996). Several APG and AUT genes have been identified, and most encode novel proteins (APG1/CVT10/AUT3, APG5, APG6, APG7/CVT2, APG12, APG13, APG14, AUT1, AUT2, andAUT7) (Kametaka et al., 1996; Funakoshi et al., 1997; Matsuura et al., 1997; Schlumpbergeret al., 1997; Straub et al., 1997; Kametakaet al., 1998; Lang et al., 1998; Mizushimaet al., 1998a). Biochemical characterization of the gene products has revealed some molecular aspects of autophagy: Apg1p/Aut3p is a novel protein kinase (Matsuura et al., 1997; Straubet al., 1997). Apg6p/Vps30p and Apg14p form a protein complex (Kametaka et al., 1998). Aut2p interacts with Aut7p/Apg8p, a homologue of rat microtubule-associated protein light chain 3 (Lang et al., 1998).

Recently, we found that the Apg12p–Apg5p conjugation system is essential for autophagy (Mizushima et al., 1998a). Apg12p has no significant homology to ubiquitin or ubiquitin-related modifiers; however, Apg12p is conjugated to Apg5p through an isopeptide bond between the C-terminal Gly residue of Apg12p and the Lys¹⁴⁹ residue of Apg5p. Ubiquitination is a posttranslational modification to present the degradation signal for proteolytic attack by 26S proteasomes (reviewed by Finley and Chau, 1991; Hershko and Ciechanover, 1992; Hershko, 1996; Hochstrasser, 1996a,b; Haas and Siepmann, 1997; Varshavsky, 1997). Ubiquitin is activated by a ubiquitin-activating enzyme (E1) in an ATP-dependent manner in which a thioester bond is formed between the C terminus of ubiquitin and a Cys residue within the E1 enzyme (Ciechanover et al., 1982; Haas et al., 1982). The ubiquitin is transferred from the E1 enzyme to a Cys residue within a ubiquitin-conjugating enzyme (E2). Finally, ubiquitin is covalently attached to a target protein by an isopeptide linkage directly from E2 or by a ubiquitin-protein ligase (E3) (Hershko et al., 1983;Bartel et al., 1990; Scheffner et al., 1993,1995; Peters et al., 1996; Zachariae et al., 1996).

Recent discoveries have revealed that the ubiquitin-related modifiers other than ubiquitin play essential roles in eukaryotes (reviewed byJohnson and Hochstrasser, 1997; Saitoh et al., 1997; Dolan, 1998; Hochstrasser, 1998). A mammalian ubiquitin-related protein, SUMO-1 [small ubiquitin-related modifier; also called GMP1, PIC1, UBL1, or sentrin (Boddy et al., 1996; Matunis et al., 1996; Okura et al., 1996; Shen et al., 1996)] is covalently attached to the RanGAP1 and PML proteins (Matuniset al., 1996; Mahajan et al., 1997; Mulleret al., 1998). This posttranslational modification affects the subcellular localization of these proteins (Mahajan et al., 1998; Matunis and Blobel, 1998; Muller et al., 1998). A yeast SUMO-1 homologue, Smt3p, is activated by an E1-like heterodimer Aos1p/Uba2p (Dohmen et al., 1995; Johnsonet al., 1997). The E2 enzyme for Smt3p and SUMO-1 are Ubc9p and its mammalian homologue (Gong et al., 1997; Johnson and Blobel, 1997; Lee et al., 1998; Schwarz et al., 1998). Another family of ubiquitin-related proteins is RUB1 and NEDD8 (Kumar et al., 1993; Callis et al., 1995;Hochstrasser, 1996; Kamitani et al., 1997; Lammer et al., 1998; Liakopoulos et al., 1998). A major substrate of RUB1/NEDD8 is Cdc53p/Cullin in yeast and mammalian cells, which play an essential role in regulating the cell cycle (Lammer et al., 1998; Liakopoulos et al., 1998; Osaka et al., 1998). In Arabidopsis thaliana, the auxin response depends on the RUB1 modification of nuclear proteins (del Pozo et al., 1998). RUB1 and NEDD8 are activated by E1-like heterodimers: Ula1p (Enr2p)/Uba3p in yeast, an APP-BP1 (a 59-kDa β-amyloid-protein-precursors-binding protein)/human UBA3 homologue in humans, and AXR1/ECR1 in A. thaliana (Leyser et al., 1993; Chow et al., 1996; del Pozo et al., 1998; Liakopoulos et al., 1998; Osaka et al., 1998). The E2 enzyme for RUB1 and NEDD8 are Ubc12p and its mammalian homologue (del Pozo et al., 1998; Liakopouloset al., 1998; Osaka et al., 1998).

These findings strongly suggest that there must be E1- and E2-like enzymes for the Apg12p–Apg5p conjugation system in yeast. Candidates in the Apg12p conjugation system are Apg7p/Cvt2p and Apg10p. Inapg7 and apg10 mutants, no Apg12p–Apg5p conjugate is observed, indicating that Apg7p and Apg10p play indispensable roles in the Apg12p conjugation system (Mizushimaet al., 1998a). A region of Apg7p (residues 322 to 407 out of 633 amino acids) shows significant homology to the corresponding region of a ubiquitin-activating enzyme, Uba1p, although the other regions show no homology (McGrath et al., 1991; Mizushimaet al., 1998a). We took particular interest in Apg7p/Cvt2p and investigated functions of Apg7p through both biochemical and molecular biological techniques. In this study, we provide several lines of evidence showing that Apg7p is an Apg12p-activating enzyme.

MATERIALS AND METHODS

Strains, Media, Materials, and Genetic Techniques

Escherichia coli strain DH5α as the host for plasmids and protein expression was grown in Luria Broth medium in the presence of antibiotics as required (Ausubel et al., 1995). The S. cerevisiae strains and plasmids used in this study are listed in Table 1. All yeast strains were cultured in rich medium (YPD, pH 5.0: 1% yeast extract, 2% polypeptone, 2% glucose, 20 mg/l adenine, 20 mg/l tryptophan, 20 mg/l uracil, and 50 mM succinate/NaOH, pH 5.0), MVD medium (0.67% yeast nitrogen base without amino acids, 0.5% casamino acids, and 2% glucose), or SD medium (0.67% yeast nitrogen base without amino acids, 2% glucose, and appropriate amino acids as described by Kaiseret al. [1994]). Nitrogen-starvation medium contained 0.17% yeast nitrogen base without amino acids and ammonium sulfate and 2% glucose. Solid medium contained 2% Bacto agar. Standard genetic and molecular biological techniques were performed as described byKaiser et al. (1994) and Ausubel et al. (1995). The PCR was performed with a program temperature control system PC-701 (ASTEC, Fukuoka, Japan). The DNA sequence was determined by an ABI 373A DNA sequencer (PE Applied Biosystems, Foster City, CA). Restriction enzymes were purchased from TOYOBO (Osaka, Japan) and New England Biolabs (Beverly, MA). Oligonucleotides were synthesized by Sawady Technology (Tokyo, Japan) or ESPEC oligo-service (Ibaraki, Japan). pYO324 was a kind gift from Y. Ohya, pRS series vectors were kind gifts from P. Hieter, and pGAD-C1 vector, pGBD-C1 vector, and PJ69-4A strain were kind gifts from P. James (Sikorski and Hieter, 1989; Jameset al., 1996; Homma et al. 1998). pGEX-3X and cyanogen bromide-activated Sepharose beads were purchased from Amersham Pharmacia Biotech (Uppsala, Sweden); pGEM-T was from Promega (Madison, WI).

Gene Disruption of the APG7 Gene

Gene disruption with the PCR product was performed as described by Lorenz et al. (1995). Briefly, PCR was performed with APG7-pRSF primer (5′-GTCGTCAGAAAGGGTCTTAAGTTATGCACCAGCTTTTAAATCATTTCTGGTATCACGAGGCCCTTTCGTC-3′), APG7pRSR primer (5′-GCAATCTCATCAGATTCATCATCTTCCCATTCAAAAACATCGTTGCCTAGGTGCGGTATTTCACACCGC-3′), and pRS30X plasmids as a template. The amplified PCR product was transformed into a yeast strain. APG7 gene disruption was confirmed by PCR with APG7Bgl2ATG primer (5′-AGATCTATGTCGTCAGAAAGGGTCTTAAG-3′) and APG7SalISTOP primer (5′-GTCGACTATTAAGCAATCTCATCAGATTCATC-3′).

Plasmid Construction and Site-directed Mutagenesis

A BamHI fragment (3.9 kb) containing theAPG7 gene (ORF YHR171w) was introduced into theBamHI site of pRS314 and pYO324 (pAPG7-314 and pAPG7-324, respectively). To construct an expression plasmid for c-myc–tagged Apg7p, SacI and ApaI sites were introduced just before the termination codon of the APG7 gene by nested PCR with APG7SAF primer (5′-GAATCTGATGAGATTGC-TGAGCTCGGGCCCTAATATTTTGCATATAATAGC-3′), APG7SAR primer (5′-GCTATTATATGCAAAATATTAGGGCCCGAGCTCAG-CAATCTCATCAGATTC-3′), M13-47 primer (5′-CGCCAGGGTTT-TCCCAGTCACGAC-3′), and M13-RV primer (5′-GAGCGGATAA-CAATTTCACACAGG-3′). The amplified DNA fragment was cloned into pGEM-T Vector (pAPG7SA-GEMT). AStuI–SacI fragment (0.12 kb) of pAPG7-GEMT, aSacI–ApaI fragment (0.16 kb) of pMPY-3xMYC, and an ApaI–SalI fragment (1.0 kb) of pAPG7-GEMT were replaced with the StuI–SalI region of pAPG7-314 and pAPG7-324 (pAPG7myc-314 and pAPG7myc-324). The junction and PCR-amplified region were confirmed by DNA sequencing. The plasmid complemented the apg7Δ mutation.

Site-directed mutagenesis of the APG7 gene (Gly³³³ and Cys⁵⁰⁷ changed to Ala) was performed by nested PCR with APG7E1928F primer (5′-CTTTAAAAATTGCTGACCAATCCGTGG-3′), APG7X2298F primer (5′-GAGCATTAATAAAAGAGCATG-3′), APG7G333AR primer (5′-CAACCTAGTGTAGCAGCACCTAGTAGTAG-3′), APG7C507AR primer (5′-CTAGTTACTGTGGCCATTTGATCC-3′), APG7S2855R primer (5′-CCTGCTTTATGACTGACAAACCGC-3′), and pAPG7-314 as a template. The amplified EcoRI–StuI fragment (0.8 kb) was replaced with the same region of pAPG7myc-314. The mutation site and PCR-amplified region were confirmed by DNA sequencing. The resultant plasmids were designated pAPG7G333Amyc-314 and pAPG7C507Amyc-314.

For the two-hybrid experiment, a BglII site was introduced just before the start codon of the APG7 gene by PCR with APG7Bgl2ATG primer, APG7SalISTOP primer, and pAPG7-314 as a template. The PCR product was cloned into pGEM-T. TheAflII–SalI fragment (∼1.9 kb) of the clone was replaced with the AflII–SalI fragment (∼3.8 kb) of pAPG7myc-314 (pAPG7BS-GEMT). TheBglII–SalI fragment of pAPG7BS-GEMT was subcloned into the BamHI–SalI sites of the pGBD-C1 vector (pGBD-APG7). The AflII–SalI fragment of pGBD-APG7 was replaced with theAflII–SalI fragment from pAPG7G333Amyc-314 or pAPG7C507Amyc-314 to construct pGBD-APG7G333A and pGBD-APG7C507A, respectively. The whole APG12 coding region was PCR-amplified and then introduced into the PstI site of pGAD-C1 (pGAD-APG12).

Antibodies

To express the GST-Apg7p fusion protein in E. coli, an EcoRV fragment (0.6 kb) and anEcoRV–EcoRI fragment (1.6 kb) of pAPG7-316 were subcloned into the SmaI andSmaI–EcoRI sites of pGEX-3X, respectively (pGEX-APG7N and pGEX-APG7C). GST-Apg7p fusion proteins were expressed in DH5α cells carrying pGEX-APG7N and pGEX-APG7C according to manufacturer’s protocol (Amersham Pharmacia Biotech). The GST–C-terminal region of Apg7p was expressed in cells carrying pGEX-APG7C and purified on glutathione Sepharose 4B. The GST–N-terminal region of Apg7p was expressed in cells carrying pGEX-APG7N but was included in the inclusion bodies. The protein was extracted from the inclusion bodies with extraction buffer (8 M urea, 50 mM Tris-Cl, pH 9.5), subjected to SDS-PAGE, and excised from the gel. Polyclonal antibodies against the carboxyl and amino terminal fragments of Apg7p were raised in Japanese white rabbits using purified proteins as antigens (αApg7C and αApg7N, respectively). The IgG fraction was precipitated with 50% saturated ammonium sulfate and dissolved in TBS buffer (150 mM NaCl, 20 mM Tris-Cl, pH 7.5). Anti-Apg7p antibodies were purified on a GST-Apg7p Sepharose column. A polyclonal anti-API antibody was a kind gift from D. J. Klionsky. An anti-hemagglutinin (HA) mouse mAb (16B12) was purchased from Berkeley Antibody Company (Berkeley, CA), an anti–c-myc mouse mAb (9E10)-Agarose conjugate was from Santa Cruz Biotechnology (Santa Cruz, CA), and anti-yeast 3-phosphoglycerate kinase and anti-yeast Dol-P-Man synthase mouse mAbs were from Molecular Probes (Eugene, OR).

Immunoprecipitation

Cells (OD₆₀₀ = 10) grown to early logarithmic phase in MVD or YPD pH 5.0 medium were harvested and converted to spheroplasts in spheroplasting solution (1% yeast extract, 2% polypeptone, 0.5% glucose, 1 mg/ml Zymolyase 100T). The spheroplasts were harvested in 1.3 M sorbitol as a cushion, lysed with lysis buffer (1% SDS, 150 mM NaCl, 20 mM sodium phosphate, pH 7.5), vortexed, boiled for 5 min, and chilled on ice. When indicated, 1 mM DTT was added to the lysate before boiling. Ten volumes of IP buffer (2% Triton X-100, 150 mM NaCl, 20 mM sodium phosphate, pH 7.5) were added to the cell lysate, and the mixture was centrifuged at 10,000 ×g for 5 min at 4°C to remove debris. The supernatant was precleared with 50 μl Protein A-Agarose (20% slurry, Santa Cruz Biotechnology). Fifty microliters of Agarose beads conjugated with anti–c-myc antibody (9E10) (20% slurry, Santa Cruz Biotechnology) were added to the lysate, and the mixture was incubated for 2 h at 4°C. The immunoprecipitate–bead complex was washed six times with ice-cold RIPA buffer (10 mM Tris-Cl, pH 7.4, 1% Nonidet P40, 0.1% sodium deoxycholate, 0.1% SDS, 150 mM NaCl, 1 mM EDTA). Proteins were eluted from the beads with 0.1N glycine-HCl, pH 2.5, and precipitated by incubation with 10% TCA on ice for 1 h. The sediment was washed twice in cold acetone, subjected to reducing SDS-PAGE, transferred to a polyvinylidene difluoride membrane (Millipore, Bedford, MA), and analyzed by Western blotting. All solutions contained a protease-inhibitor mixture for use with fungal and yeast extracts (Sigma, St. Louis, MO).

Other Techniques

Two-hybrid analysis was performed as described by James et al. (1996). A biochemical assay monitoring autophagy in yeast by alkaline phosphatase processing was performed as described by Nodaet al. (1998). Subcellular fractionation of yeast cells was performed as described by Huang and Chiang (1997). Advanced BLAST search was performed on the National Center for Biotechnology Information web site (http://www.ncbi.nlm.nih.gov/BLAST/). Multiple alignment of amino acids sequences was performed using a CLUSTAL W program by the computer laboratory in the National Institute for Basic Biology.

RESULTS

Apg12p Interacts with Apg7p via a Thioester Bond

A significant homology between Apg7p and Uba1p suggests that Apg7p functions as an Apg12p-activating enzyme. In this case, Apg12p should first form a transient enzyme–substrate complex with Apg7p as demonstrated with Uba1p and ubiquitin. To examine this possibility, we constructed a strain of yeast expressing both c-myc–tagged Apg7p and HA-tagged Apg12p (YIT702 strain), and immunoprecipitated the c-myc–tagged Apg7p with anti–c-myc antibody. Western analysis using αApg7N antibody recognized c-myc–tagged Apg7p (∼78 kDa) in the immunoprecipitates (Figure 1A, WB αApg7N). HA-tagged Apg12p was also detected in the same immunoprecipitate with anti-HA antibody (Figure 1A, WB αHA). The results correspond to those of the two-hybrid experiment (Figure2). PJ69-4A strain (trp1-901 leu2-3112 gal4Δ gal80Δ GAL2-ADE2 GAL1-HIS3) expressing both GAL4AD-Apg12p and GAL4BD-Apg7p grew well on SD-Ade-His-Trp-Leu plate, whereas strains expressing GAL4AD and GAL4BD, GAL4AD-Apg12p and GAL4BD, or GAL4AD and GAL4BD-Apg7p did not. These results suggest that Apg12p interacts with Apg7p.

We next determined whether this interaction is mediated by a thioester bond. When the lysate was treated with 1 mM DTT before immunoprecipitation, no Apg12p coimmunoprecipitated with the c-myc–tagged Apg7p (Figure 1B, DTT+). In contrast, Apg12p coimmunoprecipitated with c-myc–tagged Apg7p when the lysate was not treated with DTT (Figure 1B, DTT−). Furthermore, when a lysate of cells overexpressing both Apg7p and Apg12p (YIT703 strain) was analyzed on a nonreducing gel, two bands of Apg7p were detected by Western analysis using αApg7N antibody (Figure 1C, pAPG7myc-324/pAPG12HA-426, βME−). When a sample was treated with a thiol-reducing reagent, β-mercaptoethanol, the upper band disappeared, indicating that the upper band is a thioester conjugate of Apg7p. The conjugates were hardly detected in a lysate of cells expressing Apg7p and Apg12p on centromere-type plasmids (Figure 1C, pAPG7myc-314/pAPG12HA-316, YIT702 strain). This is probably due to the low amount of the conjugate form of Apg7p in the strain. These results indicate that Apg12p binds with Apg7p via a thioester bond.

The ATP-binding Domain of Apg7p Is Essential for the Interaction with Apg12p to Form the Apg5p–Apg12p Conjugate

Apg12p–Apg5p conjugation was reconstituted in an ATP-dependent manner in vitro (Mizushima et al. 1998a). A Uba1p-homologous region in Apg7p contains a putative ATP-binding domain (GXGXXG; residues 331 to 336) (Figure 3A) (Wierenga and Hol, 1983; McGrath et al., 1991). According toWierenga and Hol (1983), the second Gly of the ATP-binding domain will be essential for function. To investigate whether the ATP-binding domain is essential for Apg7p function, we used the site-directed mutagenesis to change the Gly³³³ of Apg7p to Ala (Figure3A). The mutant protein was expressed in an amount quite similar to wild-type Apg7p (Figure 3B); however, it could not precipitate HA-tagged Apg12p (Figure 3C). This result indicates that the ATP-binding domain of Apg7p is essential for the conjugation of Apg12p with Apg7p.

If Apg7p functions as an activating enzyme for Apg12p–Apg5p conjugation, the loss of conjugation of Apg7p^G333A with Apg12p may result in a defect in the formation of the Apg12p–Apg5p conjugate. In wild-type cells expressing HA-tagged Apg12p, the HA-tagged Apg12p–Apg5p conjugate was recognized, whereas no conjugate was recognized in apg7Δ cells expressing HA-tagged Apg12p, as described previously (Figure 4, pAPG7myc-314/pAPG12HA-316 and pRS314/pAPG12HA-316) (Mizushima et al., 1998a). In cells expressing Apg7p^G333A, very little conjugate was recognized (Figure 4, pAPG7G333Amyc-314/pAPG12HA-316). These results suggest the ATP-binding domain of Apg7p is essential for the formation of the Apg12p–Apg5p conjugate.

Cys⁵⁰⁷ of Apg7p Is an Active Center Cysteine Essential for Its E1 Function

The active center cysteine (Cys⁶⁰⁰) of yeast Uba1p is essential for the function of a ubiquitin-activating enzyme. If Apg7p is a protein-activating enzyme for Apg12p, a Cys residue of Apg7p will be essential for Apg12p–Apg7p conjugation. The active center cysteine of Apg7p is strongly suggested to be Cys⁵⁰⁷, because the amino acid sequence of a neighboring region of Apg7p (MCTV) is identical to the corresponding regions in wheat UBA1 (MCTV; GenBank accession number [Ac.No.] P20973) and A. thaliana UBA1 (MCTV; GenBank Ac.No. U80808), and homologous to the corresponding regions in yeast Uba1p (LCTL; GenBank Ac.No. X55386), human UBA1 (ICTL; GenBank Ac.No. M58028), and mouse UBA1 (ICTL; GenBank Ac.No. D10576) (Hatfield et al., 1990, 1997; Hatfield and Vierstra, 1992;Handley et al., 1991; McGrath et al., 1991; Imaiet al., 1992).

To investigate whether the Cys residue is essential for Apg7p function, we changed the Cys⁵⁰⁷ of Apg7p to Ala by site-directed mutagenesis and expressed both c-myc–tagged Apg7p^C507A and HA-tagged Apg12p in the apg7Δ mutant with centromere-type plasmids (YIT7C507A strain) (Figure 3A). Immunoprecipitation using anti–c-myc antibody showed that no HA-tagged Apg12p was coimmunoprecipitated with c-myc–tagged Apg7p^C507A (Figure3C), although Apg7p^C507A was stably expressed in the YIT7C507A strain (Figure 3B). We further examined the formation of the Apg12p–Apg5p conjugate. No HA-tagged Apg12p–Apg5p conjugate was recognized in the lysate of the YIT7C507A strain (Figure4, pAPG7C507Amyc-314/pAPG12HA-316). These results indicate that Cys⁵⁰⁷ of Apg7p is an active site cysteine essential for the Apg12p-activating enzyme to form the Apg12p–Apg5p conjugate.

The ATP-binding Domain and Active Site Cysteine of Apg7p Are Essential for Autophagy

If the formation of the Apg12p–Apg5p conjugate via Apg7p is essential for autophagy in yeast, cells expressing Apg7p^G333A and Apg7p^C507A will show defects in autophagy. We first examined the accumulation of autophagic bodies in cells under starvation conditions in the presence of PMSF. Nomarski optics showed that no autophagic bodies accumulated in YIT7G333A and YIT7C507A cells cultured in nitrogen-starvation medium in the presence of PMSF similar to the result seen with apg7Δ cells (Figure 5A, b–d). In contrast, many autophagic bodies were seen in apg7Δ cells expressing wild-type Apg7p under the same conditions (Figure 5A, a).

We next investigated Apg7p^G333A and Apg7p^C507Afor defects in autophagy using a biochemical assay monitoring autophagy-dependent alkaline phosphatase processing (Noda and Ohsumi, 1998). The principle of this assay is as follows. A modified version of vacuolar alkaline phosphatase, Pho8Δ60p, remains in the cytosol as a proform (inactive form). When autophagy is enhanced, Pho8Δ60p in the cytosol is transferred to the vacuole and processed to the active form. This processing is monitored by assaying the activity of alkaline phosphatase in the cell lysate.

pAPG7G333Amyc-314 and pAPG7C507Amyc-314 were introduced into anapg7Δ tester strain (YTS2; apg7Δ pho8::pho8Δ60). Alkaline phosphatase activity was measured in the transformants under rich or nitrogen-starvation conditions. Under rich conditions, the alkaline phosphatase activities in cells expressing Apg7p^G333A and Apg7p^C507Awere as low as that of wild-type Apg7p, indicating that autophagy was suppressed (Figure 5B, +N). When apg7Δ cells expressing wild-type Apg7p were incubated in nitrogen-starvation medium for 4 h at 30°C, the activity increased drastically (Figure 5B, −N), indicating that autophagy is induced by nitrogen starvation; however, the activities in apg7Δ cells expressing Apg7p^G333A and Apg7p^C507A were not enhanced, indicating that autophagy is suppressed in the mutant cells.

A defect in autophagy results in a loss of cell viability under nutrient starvation conditions (Tsukada and Ohsumi, 1993). We further examined the loss of viability of YIT7G333A and YIT7C507A strains using phloxine B. Phloxine B specifically stains dead cells. The colonies of YIT7G333A and YIT7C507A cells turned red (gray in monochrome) on nitrogen-starvation plates containing 10 μg/ml phloxine B, whereas the color of apg7Δ mutant colonies expressing wild-type Apg7p was pink (white in monochrome) (Figure 5C). The results indicate that the viability of YIT7G333A and YIT7C507A cells was markedly decreased under nitrogen-starvation conditions. From these results, we conclude that the ATP-binding domain and Cys⁵⁰⁷ are essential for the function of Apg7p in autophagy.

The ATP-binding Domain and Active Site Cysteine of Apg7p Are Essential for Cytoplasm-to-Vacuole Targeting of API

The apg7 mutant also has a defect in the cytoplasm-to-vacuole targeting of API. API is synthesized in the cytoplasm as a precursor form (Klionsky et al., 1992). After targeting to the vacuole, the precursor is processed to the mature form in the vacuole (Klionsky et al., 1992). We examined whether Gly³³³ and Cys⁵⁰⁷ in Apg7p are also essential for the targeting. We prepared lysates of cells in logarithmic phase and detected the precursor and mature forms of API by Western blotting using αAPI antibody. In wild-type cells, the mature form of API was detected in addition to the precursor, indicating that the precursor transferred from the cytoplasm to the vacuole to be processed into the mature form (Figure 6, pAPG7myc-314/pRS316 and pAPG7myc-314/pAPG12HA-316). In contrast, the processing of API was markedly decreased in cells expressing Apg7p^G333A and completely inhibited in cells expressing Apg7p^C507A, as in the case of apg7Δ cells (Figure 6, pAPG7G333Amyc-314/pAPG12HA-316, pAPG7C507Amyc-314/pAPG12HA-316, and pRS314/pAPG12HA-316). These results indicate that the ATP-binding domain and Cys⁵⁰⁷ of Apg7p are essential for the cytoplasm-to-vacuole targeting of API under rich conditions.

Apg7p Is Present in Cytosol

It has been shown that most free Apg5p, Apg5p–Apg12p conjugate, and more than half of Apg12p are present in the 100,000 ×g pellet (Mizushima et al., 1998a). To determine the intracellular localization of Apg7p, a lysate of wild-type cells was fractionated by centrifugation as described previously (Huang and Chiang, 1997). Apg7p in each fraction was detected by Western blotting using αApg7C antibody. Apg7p in vegetatively growing cells was present mainly in the 100,000 × g supernatant (Figure7, wild type, Vegetative Growth). Under nitrogen-starvation conditions, the amount and localization of Apg7p remained unchanged (Figure 7, wild type, Nitrogen Starvation). Similar results were obtained with the immunoprecipitation of fractions prepared from cells expressing c-myc–tagged Apg7p. These results indicate that Apg7p is mainly present in the cytoplasm.

DISCUSSION

In this article, we have characterized Apg7p as an Apg12p-activating enzyme and demonstrated its indispensable role in the yeast autophagy. Apg12p, a novel modifier protein, binds to Apg7p via a thioester bond. This binding requires both the active site cysteine (Cys⁵⁰⁷) and the ATP-binding domain. Accordingly, both the cysteine residue and the ATP-binding domain are essential for Apg12p–Apg5p conjugation and for autophagy as well. From these results, we conclude that Apg7p is a novel E1-like enzyme that activates Apg12p and is essential for autophagy.

Although we clearly showed the E1-like function of Apg7p, only the C-terminal region of Apg7p (residues 322 to 407) shows homology to Uba1p. According to the similarity boxes within other E1-like enzymes as described by Johnson et al. (1997) and Liakopouloset al. (1998), Apg7p has an ATP-binding domain within box I and an active site cysteine within box III that are essential for E1 function (Figure 8A); however, box III of Apg7p is not similar to those of Uba1p, Uba2p, or Uba3p. Also, neither similarity box II nor IV is present within Apg7p. Phylogenetic analysis of the regions containing box I and box III in the E1 enzymes also suggested that the relationship of Apg7p to the corresponding regions in UBA1, UBA2, and UBA3 family proteins is distant (Figure 8B) (Hatfield et al., 1990, 1997; Handley et al., 1991; McGrath et al., 1991; Imai et al., 1992;Dohmen et al., 1995; Liakopoulos et al., 1998;Mizushima et al., 1998a; Osaka et al., 1998; Yuanet al., 1998). The comparison of the similarity boxes and phylogenetic analysis showed that Apg7p is distinct from other E1-like enzymes. It is of interest to know whether Apg7p functions as a heterodimer, because the Aos1p/Uba2p and Ula1p/Uba3p complexes function as heterodimers (del Olmo et al., 1997; Johnson et al., 1997; Liakopoulos et al., 1998). At present, we have no answer.

A working hypothesis of the Apg12p conjugation system consistent with these results is proposed in Figure 9. The C-terminal Gly of Apg12p is activated by Apg7p in an ATP-dependent manner. The Apg12p–Apg7p conjugate forms via the C-terminal Gly of Apg12p and Cys⁵⁰⁷ of Apg7p. The reaction probably occurs in the cytosol or the cytoplasmic side of an Apg12p-associated compartment(s), because Apg7p is mainly present in the cytoplasm (Figure 7) (Kim et al., 1999). By analogy to ubiquitination and related modifications, there may be novel E2 and E3 enzymes for the conjugation system. One such candidate for E2 is Apg10p, because the Apg12p–Apg5p conjugate is not observed in anapg10 mutant. Cloning of the APG10 gene and biochemical analysis will reveal the function of Apg10p. Screening other candidates will be performed using an Apg12p-affinity column, an Apg7p-affinity column, or two-hybrid screening.

It remains to be determined at which step in autophagy the Apg12p conjugation system works. Apg7p is mainly present in the cytosol, whereas more than half of Apg12p is localized in membrane compartment(s) or a large complex. The Apg12p–Apg7p binding probably occurs in the cytosol or the cytoplasmic side of an Apg12p-associated structure(s). Recently, Yuan et al. (1999) suggested that a quite similar conjugation system is required for microautophagy inPichia pastoris. Microautophagy is the sequestration of cytoplasmic components (peroxisomes in this case) by an invagination of the vacuolar membrane. In a conjugation-deficient mutant (gsa7), the sequestration is not accomplished completely, probably because of defects in the membrane fusion step. Gsa7p, aP. pastoris homologue of Apg7p, was shown to be conjugated to a small protein through a thioester bond. Although a modifier and substrate(s) have not been identified, it strongly suggests that the Apg12p conjugation system is essential for microautophagy. Because the process of microautophagy is quite different from that of macroautophagy, it would be interesting if similar machinery is involved in both processes.

We have identified a human Apg12p homologue and have observed that the Apg12p homologue also conjugates with a human Apg5p homologue, which has been identified as an apoptosis-specific protein (Hammondet al. 1998; Mizushima et al. 1998b). A BLAST search of Apg7p in the Expressed Sequence Tag database showed potential mammalian homologues of Apg7p. Recently, an Apg7p homologue in P. pastoris, Gsa7p, and a human Gsa7p/Apg7p homologue were identified by Yuan et al. (1999). The human Gsa7p/Apg7p homologue is expressed in various tissues as revealed by Northern analysis (Tanida and Kominami, unpublished observations). These findings suggest that the Apg12p conjugation system generally functions in eukaryotes. By analogy to the Apg5p homologue, an Apg7p homologue may play a significant role in the autophagic and apoptotic pathways in mammalian cells. Cloning and biochemical analyses of Apg7p homologues will reveal the function of the novel protein conjugation system in mammalian cells.

Apg7p is distantly related to UBA1, UBA2, and UBA3 family proteins (Figure 8A). Nevertheless, Apg7p functions as an E1 enzyme for Apg12p–Apg5p conjugation. These findings suggest that the ubiquitin-like protein modification system will form a ubiquitous regulatory system in eukaryotes, not specific to ubiquitin and ubiquitin-like proteins. Further analyses of potential E1 enzymes may reveal a novel regulatory system by posttranslational modification.

Abbreviations used:

Ac.No.	accession number
API	aminopeptidase I
GAL4AD	GAL4activation domain
GAL4BD	GAL4 DNA binding domain
HA	hemagglutinin

REFERENCES